The quality of life cycle inventory (LCI) data is crucial to the reliability of decisions made via life cycle analysis (LCA). However, many LCI data, be they from commercial software or public domain databases, violate the laws of thermodynamics due to errors, missing data, and other inconsistencies. The process engineering method of data rectification is appealing for improving the quality of LCI data, but applying this approach poses many new challenges that are not addressed in traditional rectification. One such challenge is due to the availability of life cycle data at multiple overlapping scales. This includes engineering data at the equipment and process scales, process LCA data at the value chain scale, and economic input-output and toxic release inventory data at the economy scale. This article develops a method for rectification of such multiscale data. Rectification is accomplished via a mixed integer optimization-based approach, and it is integrated with the existing computational methods for different types of traditional and hybrid LCA. The developed method may also be used for multiscale data from process engineering. It is applied to data for a caustic soda process from a public domain LCI database to illustrate the benefits of the proposed approach and identify further challenges. (c) 2007 American Institute of Chemical Engineers.